Abstract
In order to elucidate the reliability of classification, composite-score method (mentioned in Ikeda, 1969), was analyzed in terms of amount of information. First, entropy and amount of information, which is neccessary for classification of examinees into two categories of pass and fail under a given “efficiency”, is calculated. Here, “efficiency” is defined as the conditional probability of classification in terms of true score according to any rule. Next, amount of information obtained from the classification in terms of composite-score, was calculated and compared with the first. Finally, the reliability of classification in terms of composite-score is discussed with respect to those points.
